Page 1 of 6

Lithium Battery Monitor 48 cell

Posted: Thu Jun 07, 2012 7:31 pm
by dillond666
I have been thinking a lot about the pros and cons of cell balancing. I have also been pondering the ramifications of a BMS fault scenario (fire?).
There has also been mention on the web of the BMS current draw contributing to imbalancing of the cells.

I will be buying a lithium pack soon and here is the monitoring approach I propose to take.
I will charge each cell individually to 3.9v for the first time using a lab power supply. Thereafter I propose to balance manually at two month intervals and see
how far the cells drift over that period.
For charging and discharging I propose to use a cell monitoring (not balancing) solution whereby the charger cuts off when the first cell reaches 3.7v and when discharging set an alarm when the lowest cell hits 2.9v. I plan to use the Maxim part MAX11080 which is a 12 channel battery pack fault monitor. The datasheet for this part looks good and the part is about £2.30. I think this will be a robust and EMI tolerant system and as it is daisy chainable I should be able to make a 48 way board. Power consumption is neglegible compared to a picaxe and the system only needs two isolated ports for communication with a simple picaxe based master. For the isolator I propose to use the Analog Devices ADuM1201 for its miniscule power consumption and ability to shift logic levels. The parts will be surface mount and the project another leap into the unknown :lol:

Derek

Re: Lithium Battery Monitor 48 cell

Posted: Thu Jun 07, 2012 10:52 pm
by GregsGarage
I've been working on a similar idea. I am coming to the conclusion that on-board balancing is probably not necessary and have an idea for automating individual cell balancing by charging individual cells, but more on that later.

Here is an idea I am working on. It is based on the LTC6802-1 which can measure from 4 to 12 cells with a voltage range of 10 to 60 volts for all cells being measured. This chip is stackable, but I have decided that it would be better to have each with its own pic to control it. I am still using the idea of a master and slave bus, you just stack however many you need, wherever you need them in your vehicle. Attached is the schematic, the design is in Kicad.
LTC6802-1.pdf
Version A of 12 cell slave.
(83.24 KiB) Downloaded 1667 times


I will look at the Maxim MAX11080, I haven't seen that part.

Re: Lithium Battery Monitor 48 cell

Posted: Thu Jun 07, 2012 11:32 pm
by dillond666
I had looked at the Linear part too because it has the option to balance but balked at the £22 cost :shock:
The Maxim part seems cheap and does the basics.
I'm using Kicad too and think it's pretty good.
Greg, check out the isolator I'm talking about, it might solve your 4mA to opto problem, Farnell has the parts.

Derek

Re: Lithium Battery Monitor 48 cell

Posted: Fri Jun 08, 2012 12:09 pm
by GregsGarage
dillond666 wrote:I had looked at the Linear part too because it has the option to balance but balked at the £22 cost :shock:
The Maxim part seems cheap and does the basics.

It gets cheaper in quantities. £15.60 for 10, £12.60 for 100. The Maxim part is at a very good price and I couldn't remember why I didn't consider it until I saw "Available Until Stocks Exhausted" on the Farnell site. I don't know if Farnell is dropping it or if the manufacture is discontinuing it, but I would like to come up with a open source design that can be produced and don't want to spend time on a chip that is being discontinued. But for a one off project it looks like you can't beat the price of the Maxim part.

I did look at several similair devices, working along the same idea that you had, just a simple under and over voltage warning. In the end what put me off was that you would have to set each devices UV and OV limit. Different people will want different limits depending on their cell chemistry and type of vehicle. But if we have a slave that only measures voltage and tempurature, then it can be used on any pack and most chemistries including nicad and nimh. You just set the master for what ever warning levels you want. At the moment we have one master design, Peter is working on a cut down version and I still want to do a master based on the Maximite. So we get a modular system, choose a master and however many slaves you need for your pack. 8)

Greg, check out the isolator I'm talking about, it might solve your 4mA to opto problem, Farnell has the parts
I've looked at it, first glance looks alright. What needs to be considered for the spi bus is Vreg can only supply 4ma and is always live, even when the ltc6802 is idle. This is because when the chip is idle all the spi ports need to be pulled high. All this has been considered with my layout (I hope :shock: ), but it won't be able to run anywhere near the max 1mhz that it is capable of. But we don't need that sort of speed. We only need one measurement per second (the LTC6802 can do around 50 per second) and then get the ~30 bytes of data out of it before the next measurement. We can run the spi bus really slow, like 1200 baud and have plenty of time spare. The pic can also anticipate when we are going to ask for readings and get it done in advance so the master doesn't have to wait for the slow data transfer between 6802 and pic.

Re: Lithium Battery Monitor 48 cell

Posted: Fri Jun 08, 2012 10:38 pm
by GregsGarage
I was working on an idea for a stronger non-isolated master and slave bus. Here it is, any comments, questions,suggestions etc.

Re: Lithium Battery Monitor 48 cell

Posted: Sat Jun 09, 2012 10:10 am
by dillond666
Doesn't this mean that when you pull the master bus low, you also pull the PIC power supply low?

Maybe you could utilise the (unused?) fourth wire on the bus as the master bus signal wire and leave wire one to be the common pic supply.
On your schematic it seems as if the supply decoupling (C18, 19 and 20) is hindering the requirement to transmit a good square wave for serial comms.

I think I like an isolated bus better, thinking about the ground loop potential of non isolated busses makes my head hurt :?

Not sure how your thoughts work when you are chewing over a circuit idea but I always try to think in terms of voltage levels from GND and when the circuit is in operation will all my GNDs be at the same potential? When thinking about comms wiring (transmission line?) I try to imagine the effect of capacitance and inductance on my ability to change voltage levels quickly enough to produce a good waveform. I'm sure my thoughts are different from a real electronics buff but hey, I'm sharing them anyway :lol:

Derek

Re: Lithium Battery Monitor 48 cell

Posted: Sat Jun 09, 2012 1:34 pm
by GregsGarage
Oops, connected the collector of Q2 to the wrong wire, it should go to the "unused" wire. :roll: I'll fix that and post the corrections. As the schematic is drawn at the moment terminal 1 (of P3-4) is +5v, 2 is ground, 3 is slave bus and 4 is master bus. I've just realized that if I use usb connectors for P3 and P4 then we can use off the shelf usb cables for the connections between master and all slaves. If we go to an isolated bus then this wouldn't work, not enough wires, but it could still be done, just have to add more wires.

Another idea, for the temperature sensors I can replace K1 with 2 RCA style connectors, there again you can buy ready made audio cables, chop them in half and attach thermistors to fit to the pack.

Re: Lithium Battery Monitor 48 cell

Posted: Sat Jun 09, 2012 11:30 pm
by GregsGarage
Here's the latest update. I fixed the error with Q2, fiddled with some resistors, replaced the tempurature connector with 2 RCA style jacks and replaced the Bus In and Bus Out connectors with USB-B and USB-A connectors respectively. USB cables can be purchased off the shelf from 0.2 meters to 5 meters. Although the signals aren't USB compatible, plugging in a USB device should cause no harm, although I wouldn't recommend trying to charge your phone from it. :lol: For J1, the battery cell connection, I am considering a d-sub 25 way connector. You can buy parallel printer cables, chop the end off, add a fuse and terminal at the end of each cell connection and away you go. Not sure if the insulation rating would be up to this though. Any ideas for J1.

Main concern is if we can get reliable communications over a non-isolated master and slave bus (you could use an isolated supply for the master which would mean the slaves are also isolated from the vehicle). I can layout the pcb and get some made up and just build the pic side of the board to test communications before spending money on the LTC6802. Once we are happy with it then we can work on the other side of the board.

Re: Lithium Battery Monitor 48 cell

Posted: Sun Jun 10, 2012 9:04 pm
by GregsGarage
I am going to start a new topic for my LTC6802 based slave. I felt like I was hijacking Derek's thread. :wink:

Re: Lithium Battery Monitor 48 cell

Posted: Mon Jun 11, 2012 5:14 pm
by dillond666
Hijacking in a good way :lol:
I'll be looking in on your thread and I hope you'll have a look at mine.
More options are always better :wink:

I have hit a snag as I do my schematic. If I use the shutdown pin, it also shuts down the vreg power (Vaa)so how do I wake the bugger up again?
It's probably dead simple but unfortunately so am I! I need to get power from somewhere to set a high logic level on the shutdown pin. I think I'll try
some kind of capacitor isolated effort. Also the isolator goes logic high on power disruption so if the master switches off the MAX11080 won't. Minor rethink here!

Derek